Integrated Molecular Landscape of Lesions in Cereblon and Its Pathway in Myeloma Patients: Insights into Mechanisms of Relapse from IMiD-Based Treatments

Mutations of Cereblon (CRBN) or members of its pathway (CRBN-pathway genes, CRBNPGs) have been reported to exhibit increased frequency in multiple myeloma (MM) patients (pts) after relapse from regimens containing immunomodulatory thalidomide derivatives (IMiDs)±Dex. However, it is unclear if these mutations cause complete loss-of-function (LOF) of CRBNPGs in a way that can explain these relapses, given that, in preclinical studies, MM cells with only partial CRBN LOF retain substantial IMiD responsiveness. We therefore performed analyses integrating whole exome sequencing, RNA sequencing, copy number variation (CNV) and structural variant data of pts in the MMRF CoMMpass study who were treated with IMiD-based therapies. We included all pts with ≥2 paired bone marrow samples before and after an IMiD-based treatment (lenalidomide [LEN], thalidomide [T], pomalidomide [POM]) (baseline and relapse samples) and analyzed the patterns of molecular lesions of 42 CRBNPGs (based on Sievers et al. Blood 2019 and Jones et al. Leukemia 2021). To assess for alternative causes of relapse, a set of 42 known MM driver genes (defined in Ansari-Pour et al. Blood 2023) was analyzed in parallel. 177 pts of the IA22 CoMMpass dataset fulfilled inclusion criteria; 172, 89, and 15 pts received LEN-, POM-, or T-based treatment, respectively. 30 pts (16.9%) had at least one mutated CRBNPG at any timepoint (pts_mut) and overall 24 CRBNPGs were mutated in at least one pt at any time. In 19 pts (63.3%), a CRBNPG mutation was present already at baseline; 11 pts (36.7%) developed a new CRBNPG mutation at relapse; and 2 pts (6.7%) had >1 CRBNPG mutated (2 and 5 genes, respectively). FAM83F (n=4; 12.5%) and IKZF3 (n=3; 9.4%) were the CRBNPGs most frequently affected. CRBN was mutated in 2 pts (6.7%) but without CNV loss, while full-length gene transcript was detected in both pts. In 4 cases, CRBN was proximal to chromosomal translocation breakpoints, but no mutations or absence of wild-type CRBN transcript were observed. In addition, no fusion transcripts involving a CRBNPG were detectable (at fusion fragment per million [FFPM]>5). Interestingly, when we examined an additional set of IA22 pts who did not undergo IMiD-treatment (n=14), the frequency of CRBNPG mutations observed (n=2, 14.3%; SALL4 and DDB1) was similar to the IMiD-treated pts. An increase in variant allele frequencies (VAF) of mutations for any CRBNPG occurred in 12 pts (37.5% of pts_mut), however, VAF did not exceed 0.5. A preexisting mutation was no longer detectable at relapse in 8 pts (26.7%) and 3 pts showed a decline in VAF. Assessment of other known MM “driver" genes as possible explanation of relapse revealed mutations in 24 pts (80.0% of pts_mut) at relapse: VAF of these genes showed an overall heterogeneous picture, with KRAS, BRAF, ARID1A, TP53 and IRF4 being among the most recurrently affected genes, consistent with their universal MM driver gene function. 10 of 11 pts (90.9%) who developed a new CRBNPG mutation and 10 of 12 pts (83.3%) with VAF increase showed partial response or better to IMiD treatment vs. 3 pts without clinical response (1 stable disease, 2 progressive disease). Only 2 of 19 (10.5%) pts with a CRBNPG mutation present at baseline were non-responsive to IMiD treatment. While 4 pts had a combined loss of one CRBNPG allele and mutation of another (for genes COPS8, COPS7B, DEPDC5, FAM83F), only 2 of these combined events occurred at relapse and with VAF values <0.4, suggesting that complete LOF of these genes was not present in a substantial fraction of the MM cell population at relapse. Most pts received combination treatment of IMiDs+proteasome inhibitor (PI), but we observed that preclinical CRISPR KO of CRBNPGs did not cause sensitization (or resistance) to PIs, suggesting that the genomic results for CRBNPGs in clinical samples were conceivably not skewed by PI use, e.g. via elimination of cells with complete LOF of CRBNPGs.
Overall, in this integrated analysis, a minority of MM pts relapsing from IMiD-based treatments harbor genomic defects of CRBNPGs but these do not involve complete LOF for any such gene. Notably, new/enriched mutations in other genes, beyond CRBNPGs, with known roles as MM drivers cannot be excluded as alternative explanations for these relapses. Additional and more complex, genomic or non-genomic, mechanisms may account for relapse from IMiD-based regimens.